Positionless Attitude Estimation With Integrated Star and Horizon Sensors

Abstract

With an increasing demand for nanosatellites to perform accurate pointing, such as on Earth observation missions, the availability and accuracy of the orbital position and velocity information on board the satellite's computer become indispensable, as, for instance, when the determination of the attitude is based on star sensor for an Earth pointing satellite. The conventional way of providing this information is to run an orbit propagator, which is a high-order integrator algorithm such as the Simplified General Perturbations Satellite Orbit Model 4 (SGP4), onboard the satellite, supplied with the position and velocity uploaded from the ground station on a regular basis, e.g., every one or two days, depending on the required orbit accuracy. For a satellite mission with a more stringent requirement, an onboard Global Positioning System (GPS) receiver is desirable to get higher accuracy. Additionally, the GPS receiver can be used for time synchronization in-orbit and for accurate process tag timing. This operation requires a great computational cost and high memory resources. Since most cubesats are limited by the available processing power and usually on-board computers do not use multi-core processing, they suffer long computing times in forced mode. This work presents a new autonomous method to determine satellite position without orbit propagation, GPS, or the necessity of receiving velocity and position from a ground. The proposed system relies on a set of a star Sensor and a horizon Sensor to perform autonomous geocentric attitude and orbit (if desired) determination. A ground device platform composed of two cameras to simulate the star and horizon sensors, together with two portable projectors to depict images of a star field and the Earth limb was created to validate the attitude determination algorithms. An open-source code for detecting the inertial orientation of the star sensor through the astronomical imaging of the sky was employed, whereas the horizon sensor attitude was computed using standard image processing methods. The hardware and software requirements are presented, and their performances are discussed. Results have shown that a significant improvement in attitude knowledge can be achieved with this strategy, even without onboard orbit propagation. In fact, it can be shown that some orbit ephemeris can also be computed by this arrangement as a side effect. © 2013 IEEE.